FreePBX conferences: roles, marked users and bridge evidence

Map every route into the conference

The Conferences module creates a dialable extension and a reusable destination. Sangoma documents that inbound routes, IVRs and time conditions can all send calls to it, so an internal extension and a public DID are different exposure paths even when they reach the same room. Inventory each path, the caller population and the terminal destination before changing participant options.

A phone's local three-way conference is not this server-side bridge. Preserve the exact conference number, inbound number or IVR choice, trunk, tenant and module version in the acceptance record. A route that reaches a PIN prompt is only route evidence; it does not establish authorization, admission or usable mixed audio.

Keep participant, administrator and marked-user roles separate

Participant and administrator PINs are access controls for different roles, not interchangeable meeting passwords. Use distinct values, restrict their distribution and rotate them when the audience or operator changes. A caller who passes a participant PIN is not therefore an administrator and must not inherit mute, lock or removal privileges.

At the underlying ConfBridge layer, admin, marked, wait_marked, end_marked and end_marked_any are separate user-profile controls. A marked user can act as a meeting-lifecycle signal without automatically being an administrator. Test the exact FreePBX fields and generated profiles rather than assuming the labels map one-to-one across module releases.

Prove the waiting and ending lifecycle

Decide whether ordinary participants may talk before a marked user arrives, whether they hear music on hold while alone or waiting, and what happens when the last or any marked user leaves. ConfBridge documents those as separate behaviors. A caller hearing a waiting prompt is not yet evidence that they are mixed into the active room.

Test the sequence deliberately: participant first, marked user or administrator second, another participant third, then each departure order. Preserve prompt, admission, participant count, mute state, bridge audio and final removal as separate observations. A bridge left empty after the test and a caller returned to the intended destination are separate cleanup checks.

Treat announcements, menus and DTMF as their own paths

Join and leave sounds, recorded-name introductions, participant counts, language prompts and per-user quiet behavior are not one switch at the Asterisk layer. ConfBridge also assigns DTMF actions through a menu profile. Verify which prompts play to the room, which play only to one caller and which DTMF sequences are available to ordinary and administrative users.

A SIP INFO, RFC 4733 or in-band digit reaching the PBX is not proof that the intended menu action ran. Capture the selected digit path, loaded menu, resulting action and unchanged behavior for unauthorized callers. Do not enable DTMF passthrough or wider administrator access merely to make one menu test pass.

Separate participant audio from bridge performance

For one caller with bad audio, inspect that participant's endpoint, codec, RTP path and audio device before changing the whole bridge. ConfBridge's jitter buffer adds delay; its talk-detection thresholds can clip speech or trigger false events when changed without evidence. The bridge mixing interval also trades processing work against added delay.

Pilot one internal pair, then an external leg, then the expected participant count. Record admission, bidirectional audio, double-talk, delay, mute and removal. External participants consume trunk and media resources, while the conference number itself is not a capacity measurement. Preserve the loaded Asterisk version and avoid generic threshold tuning without a captured symptom.

Own recording, retention and retirement

Conference recording is a separate policy decision. Asterisk documents recording from the first active user's entrance to the last active user's exit, with files placed in the configured monitoring directory by default. A configured recording switch is not proof that the right meeting was recorded, that every leg is audible, that the file is uniquely named or that authorized playback works.

Test one meeting that policy requires recording and one that must remain unrecorded. Verify notice or consent, final filename, storage path, both directions, stop behavior, retention and access. Rotate shared PINs, remove obsolete public routes and preserve an alternate calling path. No FreePBX module, route, PIN, ConfBridge profile, conference, recording, capacity or live call was inspected or tested for this guide.

Treat conference waiting music as a scoped class

Conference hold or waiting treatment can select a named music-on-hold category. Confirm the conference state, selected class and listener role rather than assuming the system default applies to every participant.

Diagnose one failed entry and media path before changing the room

When a caller waits, cannot join or has no mixed audio, use the FreePBX conference diagnostic. It separates route and PIN from participant, administrator and marked roles, loaded ConfBridge state, each media leg, the bridge mix, menus, prompts, waiting music and recording.

Separate the FreePBX destination from active-room inventory

The AMI ConfBridge snapshot guide treats a configured conference destination and a live Asterisk room as different evidence.

Do not treat an accepted ConfBridge control as a FreePBX outcome

The ConfBridge control diagnostic traces the action through Asterisk state before continuing to admission or audio symptoms.

Do not equate ConfBridge events with a complete caller result

The ConfBridge event diagnostic traces lifecycle visibility and talk detection before continuing to admission or audio symptoms.

Trace the generated destination into effective ConfBridge profiles

The ConfBridge menu diagnostic preserves generated dialplan, positional arguments, profile fallbacks, digit receipt, prompt assets and caller return without editing generated files first.